rdf-infographic-skill

name

rdf-infographic-skill

description

Generate interactive single-file HTML infographics with embedded Knowledge Graph Explorer, floating navigation, theme toggle, resolver-backed entity links, and optional matching Markdown companions from RDF Turtle or JSON-LD. RDF is always the source of truth. Use after rdf-kg-generator or on any RDF dataset to create visual exploration artifacts, clinical timeline views, comorbidity graphs, or stakeholder presentations. Supports healthcare-specific rendering for medical entities, patient events, and trial workflows.

RDF Infographic Skill (HTML + Markdown Companion Generator)

Transform validated RDF Knowledge Graphs into sophisticated, self-contained interactive HTML infographics and matching Markdown companions. This skill completes the end-to-end workflow started by rdf-kg-generator: RDF → visual exploration artifact.

When to Use This Skill (Harness Mode)

Activate RDF Infographic Harness Mode when the user requests:

• HTML infographic or KG Explorer from RDF / Turtle / JSON-LD / SPARQL results
• Interactive visualization of a knowledge graph (especially clinical, comorbidity, trial, or patient timeline graphs)
• "Create visual companion", "HTML explorer", "interactive infographic", or "regenerate the viz from this RDF"
• Paired HTML + Markdown + RDF artifact set from a previously generated KG

In Harness Mode, RDF is the single source of truth. All narrative (FAQ, HowTo, glossary, entity descriptions, relations) must be derived from the parsed RDF. Do not invent content.

Core Harness Contract (must be satisfied unless user explicitly opts out)

See the full enforceable rules in references/harness-contract.md. Key principles:

• RDF is the single source of truth — derive every FAQ, HowTo step, glossary term, entity card, timeline event, and narrative section from parsed RDF resources and properties. Counts and content must match exactly.
• Shared artifact stem across HTML, Markdown, and RDF outputs.
• Resolver-backed links (default uriburner describe pattern, configurable) with correct target="_blank" behavior for external links.
• Interactive KG Explorer with class filters, search, drag/zoom, Core/Full toggle, and healthcare color/timeline enhancements.
• Floating nav + theme system (draggable, persistent light/dark).
• POSH/JSON-LD pairing and attribution footer.
• Zero-failure gate — all checklist items in the reference must pass before delivery.

Load references/harness-contract.md on every run and follow every applicable rule.

Workflow (execute in order)

1. Receive input — Either a path/string containing RDF (Turtle preferred for readability, JSON-LD also supported) or a reference to a previously generated artifact from rdf-kg-generator.
2. Parse & analyze RDF — Use rdflib (or the provided scripts/analyze_rdf_for_viz.py) to load the graph and extract:
   • Main subject(s) / primary entity and high-degree nodes
   • Key classes (MedicalCondition, Drug, ClinicalTrial, PatientEvent, HowTo, FAQPage, DefinedTermSet, etc.)
   • Relations, labels, provenance, and SPARQL examples
   • Healthcare signal for specialized rendering and color coding
3. Elicit parameters (if not already clear):
   • Output formats: HTML only (default), HTML + Markdown, or both + updated RDF
   • Theme preference or resolver base URL
   • Focus areas (e.g., "emphasize patient timeline and comorbidities", "full KG Explorer", "executive summary view")
   • Healthcare mode (auto-detected if medical classes present)
4. Plan the artifact — Briefly describe planned sections (narrative summary, FAQ count, HowTo steps, glossary terms, KG Explorer node/edge counts, timeline view if applicable). Get confirmation on complex clinical graphs.
5. Generate —
   • Load and follow the detailed rules in references/harness-contract.md.
   • Build interactive single-file HTML (modern glassmorphism or clean healthcare aesthetic with blue/purple/green accents matching your infographic-generator style, responsive, accessible).
   • Embed KG Explorer using a lightweight CDN-friendly library (vis-network or equivalent). Derive node/edge data and filters directly from the parsed RDF.
   • Derive all narrative text, FAQ, HowTo steps, glossary, timeline events, and SPARQL examples strictly from RDF entities and properties (exact counts and content match required).
   • Add floating draggable nav, theme toggle with persistence, resolver links on every entity, and proper open-tab behavior.
   • If Markdown requested: produce parallel .md with identical logical structure and resolver links (see harness contract for parity rules).
6. Validate & deliver — Run internal checks (parse, link integrity, no missing required sections, KG data consistency). Deliver files with shared stem. Provide a short usage note (how to open the HTML, interact with the explorer, switch themes).
7. Optional further handoff — If user wants a polished static executive infographic (McKinsey-style), pass a structured summary of the KG entities/relations/timeline to the infographic-generator skill.

Integration with Existing Skills

• rdf-kg-generator (primary upstream): After generating a Turtle KG, say "now create the interactive HTML infographic and Markdown companion using rdf-infographic-skill". The two skills together give a complete RDF → validated Turtle → beautiful interactive explorer workflow.
• infographic-generator (downstream/static polish): Use this skill for interactive exploration; use infographic-generator when you need a clean, presentation-ready static visual summary or dashboard image for leadership review. They complement each other.
• Your GraphDB / clinical pipelines: Generated HTML can embed live resolver links back to your GraphDB or public LOD resolvers. Markdown companions are great for documentation or Git-based workflows.

Technical Notes & Constraints

• Single-file HTML preferred for easy sharing and offline use (embed JS/CSS via CDN where possible for the explorer).
• For very large graphs (> few hundred nodes), offer "Core view" (main entities + key relations) vs "Full view" with progressive loading or clustering.
• Healthcare rendering: Prioritize clear visual distinction for conditions vs drugs vs events vs trials. Support horizontal/vertical timeline rendering when :PatientEvent or schema:MedicalEvent sequences are present.
• No external dependencies at runtime for the final HTML (self-contained after generation).
• SPARQL examples in the RDF are rendered as expandable accordions with correctly encoded live query links when an endpoint is known.

Limitations & Responsible Use

• The interactive explorer is for exploration and communication, not a replacement for full GraphDB workbench or dedicated KG visualization tools on very large enterprise graphs.
• Always validate clinical accuracy of the underlying RDF before using the visual artifact in decision support or external sharing.
• Theme and interaction persistence uses localStorage; provide fallback instructions.

This skill, together with rdf-kg-generator, gives you a powerful, repeatable pipeline for turning clinical knowledge, trial protocols, comorbidity analyses, and pharmacy data into traceable, interactive, stakeholder-ready artifacts while keeping RDF as the canonical source of truth.